Skip navigation
Please use this identifier to cite or link to this item: http://arks.princeton.edu/ark:/88435/dsp019k41zh62d
Title: A Deep Learning Approach to Aircraft Wing Optimization
Authors: Andrade, Kevin
Advisors: Martinelli, Luigi
Department: Mechanical and Aerospace Engineering
Certificate Program: Robotics & Intelligent Systems Program
Class Year: 2021
Abstract: This thesis investigates neural networks in the context of aircraft wing design optimization. This problem is broken down into three phases investigated throughout this paper: a Radial Basis Function Neural Network, a Deep Neural Network with static training, and a Deep Neural Network with dynamic training. The first phase, seeks to develop a foundation and proof of concept for the accuracy that can be achieved via neural networks, with a simplified three input (Mach Number, area, and angle of attack) and single output model that predicts the induced and zero-lift drag coefficient for both subsonic and supersonic flow speeds. In the second phase of this study a more complex deep neural network is constructed that takes in seven geometric features as inputs (Mach number, area, angle of attack, span, sweep, tip-to-chord ratio, and taper ratio) and simultaneously predicts the induced and zero-lift drag coefficients across the subsonic and supersonic regimes. The final phase of this study develops a data sampling algorithm in order to minimize the number of simulations necessary to generate a database that leads to accurate predictions. Each phase is geared towards making machine learning an effective, efficient, and powerful tool for aircraft wing high-fidelity analysis and optimization.
URI: http://arks.princeton.edu/ark:/88435/dsp019k41zh62d
Type of Material: Princeton University Senior Theses
Language: en
Appears in Collections:Mechanical and Aerospace Engineering, 1924-2021

Files in This Item:
File Description SizeFormat 
ANDRADE-KEVIN-THESIS.pdf2.06 MBAdobe PDF    Request a copy


Items in Dataspace are protected by copyright, with all rights reserved, unless otherwise indicated.